Paper accepted for presentation at Artificial Economics 2009, Valladolid, 10-11 Sept. 2009

Simulation of Effects of Culture on Trade Partner

Selection

Gert Jan Hofstede, Catholijn M. Jonker, and Tim Verwaart1

Abstract The criteria that traders use to select their trade partners differ across

cultures. The rational criterion of expected profit of the next contract to be negoti-

ated dominates the decision in individualistic, egalitarian, uncertainty tolerant cul-

tures. In other cultures, criteria like personal relations, group membership, status

difference and trust may strongly influence trade partner selection. There also ex-

ist differences in the level of information about potential partners that traders re-

quire before entering into business contacts. This paper models the role of culture

at the level of individual agents, based on Hofstede’s five dimensions of culture.

The model is applied in multi-agent simulations, that are designed as a research

tool for supply chain research. The model is implemented as a random selection

process, where potential partners have unequal probabilities of being selected. The

factors influencing the probabilities are: expected profit and trust (learnt from pre-

vious contacts with potential partners or reputation), common group membership,

societal status, and personal relations. Results are presented, that indicate that

Hofstede’s model can be used to simulate the effect of culture on the formation

and maintenance of business relationships.

G.J. Hofstede

Wageningen University, Hollandseweg 1, 6706 KN Wageningen, The Netherlands, e-mail:

gertjan.hofstede@wur.nl

C.M. Jonker

Delft University of Technology, Mekelweg 4, 2628 CD Delft, The Netherlands, e-mail:

c.m.jonker@tudelft.nl

T. Verwaart

LEI Wageningen UR, Postbus 29703, 2502 LS Den Haag, The Netherlands, e-mail:

tim.verwaart@wur.nl

1 Introduction

Strategies for selecting trade partners are known to be heterogeneous among

traders operating in the same environment. For instance, Kirman (2008) describes

trade on the Marseille wholesale fish market: according to recorded transaction

data of this market, some buyers are loyal to sellers, while others persistently dis-

play shopping behaviour, moving from seller to seller. Weisbuch et al. (2000)

showed how this heterogeneous behaviour can be reproduced in a multi-agent

simulation. The approach is based on reinforcement learning of expected profit-

ability of trade relations, where the length of an agent’s memory and its sensitivity

to past experience are parameters that differentiate agent behaviour. An interesting

observation in that research is, that both loyal buyers and shopping buyers survive

in this market.

Literature on international business, e.g. Hofstede and Hofstede (2005), Trom-

penaars and Hampden-Turner (1993) suggests that the distribution of the parame-

ters introduced by Weisbuch et al. (2000) - i.e. the length of memory and loyalty

to business relations versus the drive to explore new opportunities – will be differ-

ent across different cultures. Furthermore, besides expected profit, phenomena like

trust and personal relations are relevant and are known to have different influence

on trade partner selection and network formation across cultures (G.J. Hofstede,

2007). In some societies, economic systems may be based on trust, in other socie-

ties on opportunism. Gorobets and Nooteboom (2006) showed by means of a

multi-agent simulations that both types of systems might be viable in different so-

cieties. However, in intercultural trade these differences may hamper trade rela-

tions, because trust and opportunism may be appreciated differently. Also, loyalty

may be appreciated differently across cultures.

The relation between culture and international trade has been studied at the

macro level, e.g. (Guo, 2004; Kónya, 2006). The research reported in the present

paper models the relation between culture and trade partner selection at the micro

level. The purpose is the development of multi-agent simulations that can be used

as an instrument in supply network research, in combination with human gaming

simulations (Jonker et al., 2006; Meijer et al., 2006). The simulations and the hu-

man games are based on the paradigm of transaction cost economics (Williamson,

1985, 1998), with focus on asymmetric information, opportunism and trust. The

main processes to be modeled in the agents are trade partner selection and bar-

gaining in the pre-contract phase, and the decisions to either cooperate or defect

and either trust or monitor and enforce in the post-contract phase of transactions.

The present paper focuses on the process of trade partner selection.

The computational models of the effects of culture are based on the work of G.

Hofstede (2001). Hofstede identified five dimensions of national cultures, that can

be measured by a numerical index. The dimensions are: individualism versus col-

lectivism, inequality of power, uncertainty avoidance, masculinity versus feminin-

ity, and long-term versus short-term orientation. G.J. Hofstede et al. (2006, 2008a,

2008b, 2008c, 2009) describe production rule models of the influence of culture

on trade processes for each of the individual dimensions. Section 2 of the present

paper summarises the analyses reported in these models in as far as they are rele-

vant for trade partner selection.

Although other dimensional models of culture could certainly be used for simi-

lar purposes, Hofstede’s framework was chosen over possible other candidates

(such as Hall, 1976; House et al., 2004; Schwartz, 1994; Trompenaars and Hamp-

den-Turner, 1993) for various reasons. First, Hofstede’s work is parsimonious and

accessible, with only five dimensions compared to GLOBE’s 18, and with its 1-to-

100 scales. Second, it has a wide scope, compared to Trompenaars and Hampden-

Turner, whose dimensions are statistically intercorrelated and can be described as

aspects of only individualism and power distance (Smith et al., 1996) or Hall who

focused on the dimension of individualism (low-context communication) versus

collectivism (high-context communication). Those models miss out on issues re-

lated to gender roles, anxiety and Confucian values. Third, it has the greatest em-

pirical base of these studies, with a well-matched sample of 117.000 respondents

to the original study plus hundreds of replications during a quarter century that

validate the model (Kirkman et al., 2006; Schimmack et al., 2005). Fourth, it is the

most widely used. It has survived fashions and hasty storms of criticism (Smith,

2006; Sóndergaard , 1994). Fifth and most important, it shows continued predic-

tive value for many societal phenomena (Hofstede, 2001; Smith, 2002). The most

likely candidates for extension of the Hofstede model are the new dimensions

found by Minkov using World Value Survey data (Minkov, 2007).

This paper aims to integrate the rules for the individual Hofstede dimensions

into a model of the partner selection process, simultaneously taking all five di-

mensions into account. The basis of the model is the reinforcement learning model

proposed by Weisbuch et al. (2000), enhanced with “non-rational” aspects that are

relevant from the culture perspective. Section 3 describes the model.

The main goal of the authors’ current research is to assess the feasibility of the

Hofstede dimensions for agent-based simulation of the effects of culture on inter-

national trade, in particular in international supply chains of food products, where

intensive trade among many small-scale firms occurs, and where usually product

quality information is asymmetric. Section 4 presents results of simulations that

indicate that believable simulation results can be obtained by applying the

Hofstede model. Section 5 concludes the paper.

2 Hofstede’s Dimensions and Trade Partner Selection

Behaving as a good, upstanding member of the group is at the core of the lives

of all beings that live in social groups (Wilson, 2007). Human beings are intensely

social and spend up to twenty years being taught how to act as virtuous members

of society. But how to be virtuous? Different societies have found different an-

swers to that question. In some, rationality is a prominent virtue; in others, com-

mon sense. In some, virtue consists primarily in honouring tradition; in others, it

consists more of becoming prosperous. Although traders basically attempt to

maximize profits, their cultural background sets limits to the means they use, to

the partners they deal with, to the extent they get personally involved with part-

ners, to loyalty, to the time spent on establishing relations, to bargaining tactics, to

duration of bargaining etc. (Hofstede and Hofstede, 2005; Trompenaars and

Hampden-Turner, 1993).

In a series of papers, G.J. Hofstede et al. (2006, 2008a, 2008b, 2008c, 2009)

proposed a process model of trading agents, inspired by the context of the trust

and tracing game and transaction cost economics, and described the effects of cul-

ture on the processes for each of the individual five dimensions of culture as iden-

tified by G. Hofstede (2001). The relevant processes are:

− Trade goal selection: sell or buy, what product, quality level;

− Partner selection: search for a partner to deal with, agree to start negotiation;

− Negotiation: bargain about conditions and guarantees, resulting in a contract;

− Delivery: deliver according to the contract or use opportunities to defect;

− Monitoring and enforcing: spend resources on tracing or trust the partner;

− Belief update: while dealing, record experience to apply it in the future.

The present paper focuses on partner selection. The next paragraphs summarize

the effects of culture on trade partner selection for each dimension.

Individualism versus collectivism. In individualistic societies people primarily

feel to be an individual, responsible for his or her personal actions and well-being.

Traders in individualistic societies traders actively build and maintain relations,

and cut-off in case of insufficient utility. In collectivistic societies people have

given group memberships and relations, that cannot be cut-off, and feel responsi-

ble for and loyal to their ingroup. Traders prefer ingroup partners, but outgroup

partners can get ingroup status by mutual investment in the relation.

Power distance. This dimension differentiates between hierarchical societies

where the less powerful accept that power is distributed unequally, and egalitarian

ones where power relations are functional, as in principal-agent relations. In hier-

archical societies, traders prefer business partners with equal status. They avoid

the more powerful, but cannot refuse business proposed by a more powerful.

Uncertainty avoidance. In extremely uncertainty avoiding societies, people fear

what they are unfamiliar with (xenophobia) and feel uncomfortable in uncertain

situations. Uncertainty avoiding traders are distrusting and do not deal with

strangers and people belonging to different social classes. Traders from uncer-

tainty-tolerant societies may actively search for new partners without limits.

Masculinity versus femininity. In masculine societies people are oriented to-

ward competition, performance, and material success. Traders actively search for

new partners, or better: opponents, and experience trade as a game to be won. In

feminine societies, people are oriented toward cooperation and take care for oth-

ers. They prefer relations with a good atmosphere, prefer getting acquainted be-

fore doing business, forgive betrayal but avoid repetitive cheaters.

Long-term versus short-term orientation. In long-term oriented societies, thrift

and perseverance are respected as virtues. Traders actively build and maintain

network relations and see them as an asset for future prosperity. In short-term ori-

ented societies consumption, social obligations, and face are important, for in-

stance showing off by doing business with a high status partner.

3 Representation in Agents

Data for the trade partner selection process is modeled into the agents as follows:

− the agent’s culture <IDV*, PDI

*, UAI

*, MAS

*, LTO

*>: five variables that rep-

resent the Hofstede indices, scaled to the interval [0, 1];

− parameters β and γ that represent an agent’s loyalty (β) and learning character-

istic (γ), according to the model of Weisbuch et al. (2000);

− a partner model (a set of variables) for each potential partner;

− labels that represent an agents group memberships and societal status.

An agent’s labels are visible to other agents; the other information is private.

A partner model for partner j represents an agent’s beliefs about j:

− the expected utility J’j, learnt in previous business contacts, as a basis for pref-

erence in partner selection;

− experience-based trust tj: a subjective probability that the partner will cooperate

once a contract has been closed, also representing the experienced quality of the

relation;

− group distance Dj, between partner and self, computed from group labels;

− belief about the partners societal status sj, and the status difference Sj=sj-si with

self, observed from status labels.

Note that the agents are not modeled to be aware of other agents’ cultures.

The mechanism for partner selection is based on the reinforcement learning of

expected utility proposed by Weisbuch et al. (2000). Agents select their partners at

random, with probability:

Pj = exp(βJj) / ∑j’ exp(βJj’) , (1)

where β is a parameter that represents an agent’s loyalty to partners with high val-

ues of Jj ; Jj represents the preference for a particular partner, based on experience

of profitability of previous deals with that partner, and effected by the agent’s cul-

ture. The effects of culture on partner preference are summarized in Table 1.

Table 1. Partner model information taken into account for computing preference

Culture type Trust /

relation

Distrust Ingroup Out-

group

Status

difference

Partner

status

Individualist +

Collectivist +

Hierarchical -

Egalitarian

Unc.avoiding - - -

Unc.tolerant

Masculine

Feminine +

LT-oriented +

ST-oriented +

+ indicates that the partner trait increases preference in the particular type of culture;

- indicates that the trait has a negative influence on preference.

Table 1 presents 5 factors that increase the preference for another agent, de-

pending on culture. In individualistic, feminine, or long-term oriented cultures the

quality of the trusted relation with the partner is more important than in other cul-

tures. In collectivistic cultures ingroup partners are more probable to be selected

than outgroup partners. In short-term oriented cultures, there is a special prefer-

ence for partners with a high societal status. The increasing effect of culture on

preference for Jj is computed as follows:

e+

j = max{IDV*tj, (1-MAS

*)tj, LTO

*tj, (1-IDV

*)(1-Dj), (1-LTO

*)sj} , (2)

so influence of a single factor is modeled as the product of the normalized

Hofstede index and the value of the relevant belief in the partner model, all repre-

sented on the interval [0, 1], and from these the maximal value is selected.

The decreasing effect is computed similarly:

e-j = max{UAI

*(1-tj), UAI

*Dj, PDI

*|Sj|, UAI

*|Sj|} . (3)

The total effect of culture

ej = e+

j - e-j (4)

is used to compute the agent’s preference for partner j, taking the history of previ-

ous dealing J’j and culture into account:

Jj = (1 + ej)α J’j (5)

Where the parameter α determines the extent of the cultural impact on preference.

The resulting preference Jj is used in equation (1) to compute the probability

that j will be selected. Parameter β in equation (1), representing loyalty, also de-

pends on an agent’s culture. We expect it to be increased to a maximal value in

long-term oriented societies, and to be decreased to a minimal value in uncer-

tainty-tolerant or masculine societies.

b = max{LTO*} - max{1-UAI

*, MAS

*} . (6)

Β = β’ + (βmax

-β’)(|b|+b)/2 - (β’-βmin

)(|b|-b)/2 , (7)

where β’ represents a parameter that is assigned to the agent at initialization, with

0 < βmin

< β’ < βmax

.

The experience of dealing with agent j is processed after each negotiation:

J’j(n) = (1 - γ)J’j(n – 1) + γuj(n) , (8)

where uj(n) is the utility of the n-th negotiation result with j; uj(n) = 0 if the nego-

tiation was terminated without agreement. The value of γ is expected to depend on

culture: an higher value in feminine, a lower value in uncertainty avoiding cul-

tures:

c = 1 - MAS* - UAI

* (9)

γ = γ’ + (γmax

-γ’)(|c|+c)/2 – (γ’-γmin

)(|c|-c)/2 (10)

Parameter γ’ is assigned to the agent at initialization, with 0 < γmin

< γ’ < γmax

< 1 .

After a agent has targeted a partner, applying equation (1), it sends a proposal

to negotiate about a deal. The receiver may either accept or ignore the proposal.

The proposing agents waits for some time, and if it receives no reply, it updates J’j

with uj=0 , see equation (1), and than tries and targets a partner again.

If an agent has no negotiation going on, it checks for received proposals. It may

have recent proposals from several agents simultaneously. From the simultaneous

proposers, it selects the one with the maximum preference. There is one additional

effect: agents from hierarchical societies that face a higher-ranked proposer are in-

clined to accept even if they do not prefer the partner, because it is not done to re-

fuse in that case. The acceptability is calculated for all proposers:

aj = J’j / maxj’(J’j’) + (1-J’j)PDI*max(S j, 0) . (11)

Subsequently the agent selects, from the agents that proposed to negotiate, an

agent k with maximal acceptability and decides whether to accept its proposal or

to start looking for a partner by itself, with probabilities:

p(start negotiation with k) = ak ; (12)

p(start new partner selection) = 1 - ak . (13)

4 Simulation Results

This section presents two series of simulation results. In the first series, the effects

of the individual Hofstede dimensions are investigated by varying the index of one

dimension, while keeping the other indices constant. These simulations are run in

culturally homogeneous societies, i.e. all agents having equal cultural settings and,

in some simulations, different group memberships or different societal status. The

purpose of this first series of experiments is to verify the implementation of the

model. In the second series, Hofstede’s indices for some imaginary countries are

used to simulate trade patterns emerging in multicultural settings. The results

show that believably differentiated patterns can be generated. However, the model

needs further tuning and validation with real-word data in order to generate realis-

tic results for real countries.

Tables 2 presents results of simulation runs in different cultural settings. The

simulation model is based on Meijer et al. (2006). In the simulation, agents can se-

lect partners, negotiate, deliver, and process the experience gained in these activi-

ties, to update belief about expected utility J’j and trust or quality of the relation-

ship tj . The agents are homogeneous: all agents have equal parameter settings. In

all runs, eight supplier agents and eight customer agents were trading, all with pa-

rameters α = 1 , β’ = 1.5 , βmin

= 0.3 , βmax

= 3 , γ’ = 0.3 , γmin

= 0.1 , γmax

= 0.5 .

The normalized indices of culture were all set to 0.5 , except one, which was set to

either 0.1 or 0.9. The agents had no group distance or status difference.

Table 2. Loyalty, expressed as percentage of trade contacts with the most frequently con-

tacted partner in different (artificial) cultural settings; α = 1 , β’ = 1.5 , βmin = 0.3 , βmax = 3 ,

γ’ = 0.3 , γmin = 0.1 , γmax = 0.5; all agents have status 0.5 and common group labels

Value of index PDI* UAI* IDV* MAS* LTO*

0.9 28 21 28 26 45

0.1 31 32 30 35 24

PDI* = 0.9 : hierarchical; PDI* = 0.1 : egalitarian;

UAI* = 0.9 : uncertainty avoiding; UAI* = 0.1 : uncertainty tolerant;

IDV* = 0.9 : individualistic; IDV* = 0.1 : collectivistic;

MAS* = 0.9 : masculine; MAS* = 0.1 : feminine;

LTO* = 0.9 : long-term oriented; LTO* = 0.1 : short-term oriented.

Table 3. Loyalty, with increased β’ = 3 , βmax = 10 (other setting as in Table 2)

Value of index PDI* UAI* IDV* MAS* LTO*

0.9 38 21 33 34 71

0.1 40 36 51 44 29

As may be expected from equation (6), Table 2 shows that long-term orienta-

tion, uncertainty avoidance and masculinity effect the emerging loyalty. As Table

3 shows, increasing the basic values of β’ and βmax

increases average loyalty, but

the cultural effect remains. In particular, the increasing effect of LTO* is very

strong with the high value of βmax

, because of the non-linearity of equation (1). A

similar effect occurs with low IDV* in this setting. Because of increased prefer-

ence for ingroup partners, together with increased β and the non-linearity of equa-

tion (1), the agents stick to partners they selected in the beginning of the simula-

tion. Further experiments are run with β’ = 1.5 , βmin

= 0.3 , βmax

= 3 .

In similar experiments, it was found that reducing γ’ to 0.1 reduced the learning

of loyalty so that no differentiation was found; increasing it to 0.5 did not produce

results significantly different from Table 2. In all further experiments γ’ = 0.3 .

Table 4 presents results with heterogeneous agents with respect to group dis-

tance, in homogeneous cultures. The results indicate that in uncertainty avoiding,

collectivistic, and, surprisingly, long-term oriented societies ingroup partners are

preferred; in uncertainty avoiding societies due to aversion against anything unfa-

miliar; in collectivistic societies due to ingroup preference. In the LTO society,

loyalty makes agents stick to ingroup partners they selected in the beginning

(when individual preferences are equal) because IND* = UAI* = 0.5.

Table 5 displays the effects of culture on trade situations with unequal societal

status. Trade with partners from different classes is not done in hierarchical socie-

ties. In uncertainty avoiding societies, the aversion against what is different re-

duces cross-class shopping. In the simulations with masculine agents, the agents

are less loyal, have no threshold toward contacting lower classed agents, and the

powerful agents rapidly learn exploit their power, resulting in increased cross-

class shopping.

Table 4. Outgroup shopping, expressed as percentage of trade contacts with outgroup part-

ners; settings as in table 1, except group distance: both suppliers and customers are divided

into equally sized groups 1 and 2 with group distance Dj = 1

Value of index PDI* UAI* IDV* MAS* LTO*

0.9 31 20 35 28 16

0.1 27 41 18 30 42

Table 5. Cross-class shopping, expressed as percentage of trade contacts with partners ha-

ving a different status; settings as in table 1, except status: half of suppliers and half of

customers have status 0.01 , the others have status 0.99 ; group distance Dj = 0

Value of index PDI* UAI* IDV* MAS* LTO*

0.9 24 27 34 40 35

0.1 36 35 36 31 34

The results presented so far concern artificial cultures. Table 6 presents results

obtained with cultural settings that are similar to actual average Hofstede indices

of national cultures. The results illustrate that differentiated behavior emerges with

differentiated loyalty and different inclination to outgroup shopping. Results for

China show a weak inclination to outgroup shopping. This may seam contradic-

tory with China’s position on the world market. In Chinese culture ingroup trading

is preferred, but after getting acquainted and mutual investment in the personal re-

lation, an outgroup partner may become accepted as ingroup. Once the relational

barriers are broken, uncertainty avoidance and masculinity come to effect.

The results for outgroup shopping of Sweden and USA are similar, in spite of

the different cultures. In experiments eight customer agents with USA-like con-

figuration and eight customer agents with Swedish-like configuration traded with

eight Chinese-like supplier agents. Different patterns of customer loyalty emerged,

as displayed in Tables 7 and 8. The tables display the number of successful trans-

actions between each supplier and each buyer. In the simulation with USA-like

agents, the number of empty cells is 24 on 203 transactions, and average customer

loyalty equals 46 percent. In the simulation with Swedish-like agents, the number

of empty cells is 31 on 293, and average customer loyalty equals 56 percent.

Table 6. Average loyalty and inclination to outgroup shopping in societies of agents with

two groups, with group distance Dj = 1, no status difference, other parameters as in Table 1;

the cultures are modeled with some similarity to actual national cultures

Culture

similar to

PDI* UAI* IDV* MAS* LTO* loyalty outgroup

shopping

China 0.7 0.3 0.1 0.7 0.9 68 8

India 0.7 0.5 0.5 0.5 0.7 38 22

Russia 0.9 0.9 0.3 0.3 0.3 36 15

Sweden 0.3 0.3 0.7 0.1 0.3 32 44

USA 0.5 0.5 0.9 0.7 0.3 23 42

Table 7. Number of successful transaction between 8 USA-like customer agents and 8 Chi-

nese-like supplier agents, in 500 time steps.

Agent S1 S2 S3 S4 S5 S6 S7 S8

C1 2 1 1 9 5 8

C2 1 1 5 5 1 1 25

C3 15 1

C4 6 10 6 6 4

C5 3 5 7 3 2

C6 1 6 3 5 3

C7 6 4 5 11 1

C8 10 1 3 1 10

Table 8. Number of successful transaction between 8 Swedish-like customer agents and 8

Chinese-like supplier agents, in 500 time steps.

Agent S1 S2 S3 S4 S5 S6 S7 S8

C1 1 3 6 30

C2 1 17 9 2 6

C3 4 20 1 2

C4 2 11 4 22

C5 2 18 13

C6 30 2 1 8

C7 1 11 3 15 10

C8 7 8 11 12

5 Conclusion

The contribution of this work is that it shows how a model of culture can be for-

mulated to simulate culturally differentiated behavior of agents. The model of

Hofstede (2001) has been applied to partner selection in international trade in a

context where personal relations between traders are important. The partner selec-

tion is based on the model of Weisbuch et al. (2000). Culture is modeled to effect

preference for particular partners and parameters of the partner selection mecha-

nism (the loyalty parameter and the learning parameter).

The model is implemented in agents. Multi-agent simulations have been run to

verify the correct implementation of the model and to produce example results.

Although further refinements are possible, the results show that believable behav-

iors emerge. The results qualitatively represent effects expected on the basis of

Hofstede’s theory. However, validation against empirical data in the situations that

the model aims to describe, is required to calibrate parameters to actual trader’s

behavior and to scale Hofstede’s indices to the simulation indices.

The situation that is modeled is a common market place. All agents can be

aware of all other agents. The model does not include network extensions: the

population of agents is fixed. The agents are free to select any partner, and the

partner is free to enter into negotiations or to ignore proposals. The agents have

labels that indicate their group memberships and societal status. Labels are visible

to all agents and can be used for partner selection. The information about transac-

tions is private. It is only available to the transaction partners. They can use it for

future partner selection. An important characteristic of the present model is that

agents do not have a theory of culture. They act according to their cultural pro-

gramming, but they are not aware of cultural difference with partners.

The purpose of this model of partner selection is to simulate the behavior of

players in trade games (Jonker et al., 2006; Meijer et al., 2006). In order to vali-

date the model for this purpose, it has to be integrated with models of bargaining

and contract fulfillment. The combined models can be tuned to results obtained in

human gaming simulations, and their usefulness for supply chain research can be

assessed. Those tasks remain for future research.

References

A. Gorobets A, B. Nooteboom. In C. Bruun, editor, Advances in Artificial Economics, Lec-

ture Notes in Economics and Mathematical Systems 584: 121-132. Springer, Berlin,

2006.

R. Guo. How culture influences foreign trade: evidence from the U.S. and China. Journal of

Socio-Economics 33: 785-812, 2004.

E.T. Hall. Beyond culture. Anchor, Garden City NY, 1976.

G. Hofstede. Culture’s Consequences, 2nd Edition. Sage, Thousand Oaks CA, 2001.

G. Hofstede, G.J. Hofstede. Cultures and Organizations: Software of the Mind. Third Mil-

lennium Edition. McGraw-Hill, New York, 2005.

G.J. Hofstede. Trust and Transparency in Supply Netchains: a Contradiction? In W.Y.C.

Wang et al., editor, Supply Chain Management: Issues in the New Era of Collabora-

tion and Competition. Idea Group, Hershey PA: 2007.

G.J. Hofstede, C.M. Jonker, S. Meijer, T. Verwaart. Modelling Trade and Trust across Cul-

tures. In K. Stølen, W.H. Winsborough, F. Martinelli, F. Massacci, editors, Trust Man-

agement: 4th International Conference, iTrust 2006, Proceedings, LNCS 3968: 120-

134. Springer, Berlin, 2006.

G.J. Hofstede, C.M. Jonker, T. Verwaart. Modeling Culture in Trade: Uncertainty Avoid-

ance. 2008 Agent-Directed Simulation Symposium (ADSS 2008), Spring Simulation

Multiconference 2008: 143-150. SCS, San Diego, 2008a.

G.J. Hofstede, C.M. Jonker, T. Verwaart. Individualism and Collectivism in Trade Agents.

In N.T. Nguyen, L. Borzemski, A. Grzech, M. Ali, editors, New Frontiers in Applied

Artificial Intelligence, Proceedings of IEA/AIE 2008, LNAI 5027: 492-501. Springer,

Berlin, 2008b.

G.J. Hofstede, C.M. Jonker, T. Verwaart. Long-term Orientation in Trade. In K. Schredel-

seker, F. Hauser, editors, Complexity and Artificial Markets, Lecture Notes in Eco-

nomics and Mathematical Systems 614: 107-119. Springer, Berlin, 2008c.

G.J. Hofstede, C.M. Jonker, T. Verwaart. Modelling Power Distance in Trade. In N. David,

J.S. Sichman, editors. Multi-Agent-based Simulation IX, MABS 2008, Revised Selected

Papers, LNAI 5269: 1-16. Springer, Berlin, 2009.

R.J. House, P.J. Hanges, M. Javidan, P.W. Dorfman, V. Gupta. Leadership culture and or-

ganizations: The GLOBE study of 62 societies. Sage, Thousand Oaks CA, 2004.

C.M. Jonker, S. Meijer, D. Tykhonov, T. Verwaart. Multi-agent Model of Trust in a Human

Game. In P. Mathieu, B. Beaufils, O. Brandouy, editors, Artificial Economics. Lecture

Notes in Economics and Mathematical Systems 564: 91-102. Springer, Berlin, 2006.

B.R. Kirkman, K.B. Lowe, C.B. Gibson. A quarter century of Culture's Consequences: a

review of empirical research incorporating Hofstede's cultural values framework.

Journal of International Business Studies, 37: 285-320, 2006.

A. Kirman. Artificial Markets: Rationality and Organisation. In K. Schredelseker, F.

Hauser, editors, Complexity and Artificial Markets, Lecture Notes in Economics and

Mathematical Systems 614: 195-234. Springer, Berlin, 2008.

I. Kónya. Modeling Cultural Barriers in International Trade. Review of International Eco-

nomics 14: 494-507, 2006.

S. Meijer, G.J. Hofstede, G. Beers, S.W.F. Omta. Trust and Tracing game: learning about

transactions and embeddedness in a trade network. Production Planning and Control

17: 569-583, 2006.

M. Minkov. What makes us different and similar, A new interpretation of the World Values

Survey and other cross-cultural data. Klasika i Stil, Sofia, 2007.

U. Schimmack, S. Oishi, E. Diener. Individualism: A Valid and Important Dimension of

Cultural Differences Between Nations. Personality and Social Psychology Review,

9(1), 17-31, 2005.

S.H. Schwartz. Beyond individualism / collectivism: new dimensions of values. In U. Kim,

H.C. Triandis, C. Kagitçibasi, S.C. Choi, G. Yoon, editors. Individualism and collec-

tivism: theory, application and methods, 85-119. Sage, Thousand Oaks CA, 1994.

P.B. Smith. Culture's consequences: Something old and something new. Human Relations

55(1): 119-135, 2002.

P.B. Smith. When elephants fight, the grass gets trampled: the GLOBE and Hofstede pro-

jects. Journal of International Business Studies 37: 915-921, 2006.

P.B. Smith, S. Dugan, F. Trompenaars. National culture and the values of organizational

employees: A dimensional analysis across 43 nations. Journal of Cross-Cultural Psy-

chology 27: 231-264, 1996.

M. Sóndergaard. Hofstede's consequences: a study of reviews, citations and replications.

Organization Studies 15, 447-456, 1994.

F. Trompenaars, C. Hampden-Turner. Riding the waves of culture: understanding cultural

diversity in business, 2nd edition. Economist Books, London, 1993.

G. Weisbuch, A. Kirman, D. Herreiner. Market Organisation and Trading Relationships.

Economic Journal 110: 411-436, 2000.

O.E. Williamson. The Economic Institutions of Capitalism. Free Press, New York, 1985.

O.E. Williamson. Transaction Cost Economics: how it works, where it is headed. De

Economist 146: 23-58, 1998.

D.S. Wilson. Evolution for Everyone. Delacorte Press, New York, 2007.